The roles of cloud-based systems on the cancer-related studies: a systematic literature review

The advances in Wireless-based technologies and intelligent diagnostics and forecasting such as cloud computing have significantly affected our lifestyle, observed in many fields, especially healthcare. Also, since the number of new cases of cancer has become very high, there is a need to investigate this matter deeply. Still, there is no systematic review on the application or implementation of the cloud in cancer-care services. Hence, this paper has introduced a comprehensive review of a cloud-centered healthcare system that emphasizes treatment ways in different types of cancer until Sep 2021. The results have shown that the largest study was about the relationship between cancer and the cloud associated with breast cancer. Also, the results have shown that cloud computing facilitates data protection, privacy, and medical record access. Using cloud computing in hospitals, physicians will use advanced programs and tools, and nurses will quickly access patients’ information with new Wireless-based technologies. A strong understanding of the practical aspects of cloud computing will help researchers effectively navigate the vast data ecosystems in cancer research. So, by highlighting the advantages and drawbacks of analyzed articles, this study provides a comprehensive and up-to-date report on the field of cloud-based cancer studies to fill the previous gaps.


INTRODUCTION
Recently, public health and research in this area have gotten much attention [1][2][3]. Cancer is a significant public health issue globally and is a sickness that endangering the Quality of Life (QoL) [4,5]. It is a kind of malevolent neoplasms, usually attacking the adjacent tissues [6,7]. On the other hand, Information Technology (IT) is crucial in significantly changing lifestyles [8,9]. The medical industry has also played its part and introduced many medical designs to support the personnel throughout the process, from the detection to the treatment [10][11][12]. As a new vanguard IT-based technology, cloud computing refers to presenting an IT service allowing on-demand network access to a shared category of processing resources anytime and anywhere [13][14][15]. Cloud computing has led to dramatic growth in the volume, variety, and velocity of cancer data [16]. So the main research question is whether new technologies such as cloud computing have an impact on cancer diagnosis and treatment? On the other hand, with the increased number of data and data-generating devices in healthcare settings, the health monitoring systems have started to experience issues, such as efficient processing and latency. Several health-monitoring systems have been designed using wireless sensors networks, cloud computing, fog computing, and the Internet of Things (IoT). Most of the health monitoring systems have been designed using the cloud computing architecture [17,18]. The development of computer and Internet-oriented tools has changed the nature of healthcare services using cloud computing [19][20][21]. Conventional patient-physician detection has been reached to a stronger cutting-edge notion of electronic health, in which remote online/offline therapy and detection is possible [22,23]. Therefore, the information age's appearance has produced novel chances for data gathering, computing, and representation in cancer studies [24,25]. The developments in bioinformatics, automated data recovery mechanisms, and data storage have enhanced data availability and the capability for making critical decisions based on the investigations of usually intricate and multivariable data [26]. However, the problem is still compressing and filtering this data to be readily interpretable for qualified experts, physicians, the general public, and funding institutions [27]. Also, precision medicine is the main focus of the systematic and empirical analysis of time-sensitive detections and therapies like acute cancer cases [28,29]. As cancer therapies and the likelihood of survival rates are grown, the QoL of cancer-related patients and survivors should be increasingly investigated.
On the other hand, health-associated QoL is a multidimensional block defined as physical, mental, and social performance and wellbeing [30,31]. A cloud-based medical healthcare system is a standard platform that supports the medical experts' emergency treatment over Internet communication [32,33]. The widespread acceptance of cloud-based services in the healthcare sector has resulted in a cost-effective and convenient exchange of personal health records among several participating entities of the e-health systems. Since medical records are susceptible, security protection is much necessitated [34]. The cloud-oriented platform can merge different off-theshelf and custom gadgets to maintain and improve health-associated services, practical competencies, self-confidence, and safety [35][36][37]. Also, it is in charge of explicit and implicit communication with the primary operators [30]. The incorporation of cloud services in the field of cancer guarantees the accessibility of real-time patient data constantly streamed out of different sources [38][39][40]. This, in turn, allows the ever-increasing data to be managed and shared across the healthcare network systems upon deployment into the cloud [41][42][43].
The following are some charts (Figure 1 and Figure 2) based on the latest data from cancer statistics in the world (2018 is the year that cancer information was updated in the US). According to this statistic, 1708921 new cases and 599265 deaths have been reported. In fact, there are 436 new cases of cancer and 149 deaths per 100,000 people [44]. Concerning the effective role of new technologies in the treatment of all types of cancer, authors have highlighted how cloud computing can help the cancer treatment process. The structure of this review paper is as follows. In the next section, the paper's motivation has been presented. Then the method has been presented. We have provided an overview of the previous works, the evolution of the cloud, and several types of cancer in Section 4. After that, the discussion, findings, open issues, and limitations have been presented. Finally, the conclusion has been presented. Table 11 in the appendix provides the abbreviations.

PAPER'S MOTIVATION
Various review articles have been published around cancer and the cloud. But only one review article was found to examine the role of cloud-based systems on the success of cancer. This study will highlight the advantages and disadvantages of the investigated article. In a recent study, Vetova and Borovska [6] have reviewed medical imaging designs and cloud answers for the case study of cancer detection. They have proposed the medical breast cancer imaging notion explaining the medical devices for breast cancer diagnosing and checking, the characteristics of the images they generate, and the image processing required to mining the important information for the detection. They have also studied the categorization of the medical imaging designs for breast cancer based on the used processing technology. They have investigated the standalone class of medical imaging designs and the cloud-oriented ones alongside their features. Despite this article's useful information about breast cancer imaging, the authors have focused only on one disease. Also, the effect of cloud computing has only been investigated in one way. So, this study is not comprehensive. In this article, we are going to fill these gaps and provide an extensive study on the role of cloud computing in cancer.
Since there is no SLR article reviewing the cloud in the therapeutic systems and cancer diagnosis, newly published works have not been reviewed systematically. Consequently, this study has attempted to fill this gap and provide a systematic review on the cloud-related articles in therapeutic and diagnosis systems published from Feb 1, 2012, to Sep 2021 to identify the sources for the investigation of the applications of cloud computing in cancer domains to guide a predictor analysis. In addition, challenges have been explored, and some lines have been suggested for future studies. Therefore, the research questions are presented as follows: 1. What are the prevailing therapeutic ways currently applied using the cloud in the field of cancer?
2. What are the best opportunities for the future treatment and diagnosis of cancer?
3. To what extent can we use cloud computing to prevent, treat, and detect cancer?

III. METHOD
The present paper used an SLR [3,45,46] and narrative review [47], according to the primary rules introduced by Kitchenham [

IV. OVERVIEW OF SELECTED CLOUD-BASED ARTICLES FOR CANCER TREATMENT
Many researchers have recently studied the impact of cloud-based systems on diagnostic, prevention, and treatment applications [92]. This section has reviewed the main approaches to solve cancer patients' problems in diagnosis and treatment using cloud-based systems. On-time illness detection enables physicians to manage appropriate therapy and enhance survival [93]. In the old diagnosis, skilled physicians have visually tested individual medical images for any nodule growth symptoms in the body. Nevertheless, this type of detection is usually difficult and extremely subjective because of inter-observer inconsistency and the huge bulk of the medical image data [60,94].
Scholars have done many studies in the last few decades to propose Computer-Aided Diagnosis (CAD) systems for physicians to diagnose any type of cancer, motivated by the cutting edge computing technologies like cloud computing executing intricate image processing and the ML [95,96]. For instance, producing a microbiome data analysis pipeline with amazon web services becomes possible. This pipeline is quite trustworthy, strong, and inexpensive. The mentioned pipeline helps perform microbiome data analysis [97].
The seven bridges Cancer Genomics Cloud (CGC1) allows scholars to quickly access and cooperates on huge communal cancer genomic datasets like The Cancer Genome Atlas (TCGA). It offers safe and ondemand access to data, examination tools, and calculating resources. All different scholars can effortlessly and programmatically see, demand, and search the mentioned datasets. One can quickly analyze the cloud's desired data using over 200 preinstalled, curated bioinformatics tools and workflows. Scholars can also spread the platform's scope by adding data and devices via an intuitive software development kit. By co-localizing these resources, the CGC allows scalable and reproducible analyses. Scholars can incorporate the CGC to examine cancer genomics' main questions [98]. Some used terms are explained below before investigating the selected studies (See Table 2). It is the fact or quality of happening at best possible time or at the right time to treat and diagnose disease [99].

Resolution
The meaning of resolution is the image view quality in the treatment and diagnosis of the disease [100].

System usefulness
Usefulness is the extent to which an information system will enhance a user's performance [101]. System's usefulness is shaped by the context in which it is used [102].

Ease of use
Ease of use is a fundamental concept describing how easily users can use a product. It shows the userfriendliness of the final medical product [103].

Quality
Quality is the ongoing building and sustaining relationships by assessing, anticipating, and fulfilling stated and implied needs. It is from enhancing the products, services, mechanisms, and procedures to ensure that the process is fit and effective [104].

Open data infrastruct ures
Data like statistics, maps, and real-time sensor readings are useful in decision-making, service creation, and understanding. Data infrastructure has data resources backed by people, procedures, and technology [105].

Accuracy
Accuracy is the degree to which the result of a measurement, calculation, or specification conforms to the correct value or a standard. It is the quality of being near to the real value of accuracy in treating and diagnosing the disease [106].

Efficiency
It is the capacity to prevent wasting resources, energy, efforts, money, and time in the treatment and diagnosis of the disease or in producing the desired result [107].

Cost
The cost of an object or action requires the payment of a specified sum of money before it can be acquired or done [108]. In this paper, the cost is patients' medical equipment and treatment.

Scalability
It is the quality of a mechanism to deal with an increasing bulk of work by adding assets to it [109].

Automat
Working by itself with little or no direct human control.

Early detection
The diagnosis of the disease before its progression was suggested.

Remote detection
Information mining, particularly at audio/video frequencies, is out of an electromagnetic wave.

Decision Support System (DSS)
It is an information mechanism backing corporate's decision-making [110].

A. COLON CANCER
This section reviewed the articles that investigated the cloud's role in colon cancer treatment. Colon cancer is a prevalent tumor worldwide [111]. In the last 70 years, the frequency of colon and rectum cancers has noticeably risen; they have ranked the 2nd and 3rd (in females and males, correspondingly) among the prevalent kinds of cancer in the West [112]. Although it is rare in individuals under the age of 45, its possibility grows with age. The average age of its detection is a bit above 70 [113]. The Colorectal Cancer Research (CRC) is a known neoplasm where the essential genetic modifications of illness development, from normal epithelium to metastatic disease, have been explained [114]. It is organized from thin hyperplastic epithelium to adenoma and then to carcinoma. These successive genetic alterations appear in colon carcinogenesis, proto-oncogenes, tumor suppressor genes, and Deoxyribonucleic Acid (DNA) repair genes. CRC is cancer that appears in the colon or the rectum. Physicians can also call them colon cancer or rectal cancer, considering their place of appearance. These two are usually teamed up as they have numerous shared characteristics. Cancer appears if body cells begin to grow in an uncontrolled manner. Almost all body cells can become cancer and extend to other parts [115]. This section has analyzed 4 colon cancer articles related to cloud-based systems. After reviewing the main features of each one, the authors have summarized them in Table 3. As can be seen, the researchers have tried to reduce the costs, maximize the diagnosis time, and enhance colon cancer resolution, using treatment programs and cloud-based systems.
Zhang, et al. [54] have used a network pharmacological approach to identify anti-CRC targets of Formononetin (FN) and the molecular mechanisms of FN against the CRC. They have used a tool of the DisGeNET database for the collection of CRC-based targets. A protein-protein interaction network of FN against the CRC has been obtained utilizing a STRING. All top biological functional processes and signaling pathways of FN against the CRC have been identified using a database for annotation, visualization, integrated discovery, and Omicshare cloud platform. The anti-CRC molecular mechanisms of the FN have been implicated in the suppression of cellular proliferation and regulation of cancer-related metabolic pathways.
Xia, et al. [55] have investigated the cloud-based characterization of microbial landscape in CRC and presented a novel cloud-based bioinformatics tool, iGRAMMy, that accurately, efficiently, and robustly estimates relative microbial abundance in the sequenced tumor and matched normal samples. They have identified a significantly higher abundance of overall bacterial infiltration in tumors as compared to normal samples. They have found that species like Bacteroides fragile, Bacteroides dorei, and Fusobacterium nucleatum were meaningfully higher in tumors, corroborated by the previous reports. They have identified that Bacteroides intestinalis was also significantly more abundant in tumor samples.
Also, Simjanoska, et al. [56] have examined the platform's structure for studying the CRC in the cloud. They have presented a novel design for the CRC gene expression analysis aiming at scholars and clinical individuals as the microarray tests' availability has been raised. Their structure benefits from the cloud's speed and flexibility to do significant calculations, be effortlessly upgraded with new tools, and prevent the local infrastructure's restrictions.
Yoshida [57] has proposed a new cloud supercomputing oriented detection mechanism for colon cancer, according to the laxative free virtual colonoscopy check. A high-quality mobile display system has been linked to the cloud supercomputing virtual colonoscopy to enable on-demand visualization of the whole colonic lumens and detection of colonic lesions. The initial outcomes have indicated that the cloud-supercomputing-oriented virtual colonoscopy system with motion-oriented piloting can help the workflow and effectiveness of understanding virtual colonoscopy images to screen the CRCs. • Examining the rule of cancer-related metabolic pathways.
• Advantage: Optimal biological [55] • Investigating the cloud-based characterization of microbial landscape in CRC and present a novel cloud-based bioinformatics tool, iGRAMMy.
• Advantage: High accuracy, High efficiently • Disadvantage: Low scalability [56] • Examining the platform's structure for proposing tools for the oncologists to effortlessly notice molecular goals for the therapy and conduct a quick detection.
• Advantage: High performance, Low cost of treatment, High scalability • Proposing a new cloud supercomputing oriented detection mechanism according to the laxative free virtual colonoscopy check.
• Advantage: Easy navigation, Localization of colonic lesions, High resolution, Timeliness

B. LUNG CANCER
In this section, we have reviewed the articles about the cloud's role in the treatment of lung cancer. Scholars have recognized lung cancer as a dangerous illness that raises the universal mortality rate yearly [116]. Lung cancer has been histologically categorized into small-cell and big-cell lung cancers. Its prevalent signs are cough, dyspnea, hemoptysis, and systemic signs like weight loss and anorexia [117]. High-risk patients having symptoms should take chest radiography. If there is no other detection, physicians should use computed tomography and positron emission tomography [118]. Lung cancer has numerous causes like active cigarette smoking, being in contact with cigarette smoke (passive smoking), pipe-smoking, occupational contact with agents like asbestos, nickel, chromium, and arsenic, being in contact with radiation (like radon gas in homes and mines), and being in contact with air pollution (internal and external).
Despite recognizing this collection of organized fundamental risk factors, lung cancer's universal epidemic is mainly due to a solo one: cigarette smoking. The significance of marketing and promotion of an addicting and fatal product by international companies, considering cigarette smoking supremacy, can be perceived [119]. A reduction in lung cancer rate has been reported twice as fast in men as in women, showing the historical variances in tobacco acceptance and stoppage and the increases in women smoking occurrence in some birth cohorts [120]. In the last decade, the exponential growth of cloud computing has upgraded the healthcare facilities regarding lung cancer. In this section, five lung cancer articles related to cloud-based systems were analyzed. After reviewing the main features, they were summarized in Table 4. As shown in this table, the researchers have tried to reduce the detection time, maximize the accuracy, and enhance the resolution and early detection in lung cancer using treatment programs and cloud-based systems.
Masood, et al. [58]  Valluru and Jeya [59] have presented an optimal Support Vector Machine (SVM) for lung image grouping in which the factors of SVM are improved and characteristic choosing done by the revised grey wolf optimization algorithm integrated into the Genetic Algorithm (GA). They have also done the tests in 3 aspects: factor optimization, characteristic choosing, and optimal SVM. They have used a benchmark image database with 50 low-dosage kept lung CT images to evaluate the proposed method's functionality. It has shown its dominance on all the used images regarding a few dimensions. Also, it has reached normal exactness for the categorization meaningfully better than the other ones.
Furthermore, Junior, et al. [60] have offered a public non-relational document-based cloud-centered database of pulmonary nodules having 3D texture qualities, detected and categorized into 9 diverse subjective features by skilled radiologists. They have aimed at enhancing computer-assisted lung cancer detection, pulmonary nodule detection, and categorization research, using the proposed database in a cloud database as a service outline. They have offered pulmonary nodule data by the lung image database consortium and image database resource initiative. The extended database has excellent applicability for combining many diverse CAD mechanisms. The introduced database now has 379 tests, 838 nodules, 8237 images, 4029 CT scans, and 4208 manually-segmented nodules. Moreover, this nodule database has high applicability in the big data because NoSQL has low coupling by combining it into the patients' e-archive and other databases.
Also, Sueoka-Aragane, et al. [61] have assessed a cloud-oriented local-read pattern for imaging assessments in oncology medical tests for lung cancer. They have done two studies: The KUMO I and KUMO I Extension. KUMO I was a pilot study to show the practicability of cloud execution and detecting problems around the diversity of the assessments. 2 oncologists (Japan) and one radiologist (France) have independently assessed the chest Computed Tomography (CT) scans at 3 time-points from the starting point to development, from ten patients having lung cancer treated with EGFR tyrosine kinase inhibitors, using a cloud-oriented software answer. The KUMO outcomes have indicated the discordance rates of 40 percent for the desired lesion selection, 70 percent for the total reaction at the first time-point, and 60 percent for the whole response at the second timepoint. The KUMO I extension has added a cloudoriented quality control service to reach an agreement on choosing the desired lesions, leading to an enhanced rate of agreement of reaction assessments.
The key cause of the discordance has been the variances in choosing the desired lesions.
Ma, et al. [62] have presented a model to evaluate the disease-associated sign burden and QoL in Chinese chemo-naïve acute lung cancer patients. The mentioned patients suffering from grade III/IV lung cancer have been registered. They have used the functional assessment of cancer therapy-lung scale and cloud QoL system. The outcomes have indicated that 376 qualified patients have been studied. The top 3 severe signs were appetite loss, trouble in breathing, and cough. There was a meaningful association between the QoL and signs. Regression analysis of the QoL has shown that nearly all symptoms (but the trouble in breathing) have been the QoL's negative pointer. • Main aim: Proposing a public non-relational document-based cloudcentered database of pulmonary nodules with 3D texture qualities.
• Advantage: Early detection • Main aim: Assessing a cloud-oriented local-read pattern for imaging assessments in oncology medical tests for lung cancer.
• Advantage: High imaging evaluations • Disadvantage: Low number of samples [62] • Main aim: Presenting a model to evaluate the disease-associated sign burden and QoL in Chinese chemo-naïve acute lung cancer patients.

C. OVARIAN CANCER
In this section, the articles are reviewed, which investigate the role of the cloud in the treatment of ovarian cancer. Ovarian cancer is the 5th key reason for cancer death. Regarding the mortality rate, it is the 1st among gynecologic cancers. Ovarian carcinomas are a diverse set of neoplasms conventionally subcategorized consistent with the kind and extent of differentiation. The existing medical management of ovarian carcinoma could not consider this diversity, but each primary histological type has distinctive genetic flaws deregulating certain signaling lanes in the tumors [121,122]. In 2018, clinicians had detected nearly 22,240 new instances of ovarian cancer and documented 14,070 deaths as a result of that in the United States. A human can help decrease the death rate and frequency of ovarian cancer by dropping the racial inequalities and fostering knowledge of etiology and tumorigenesis to reinforce tactics for inhibition and on-time diagnosis [123]. In this section, 4 ovarian cancer articles related to a cloud-based system were analyzed. Then, the key features of the articles were summarized in Table 5. As can be seen, the researchers have tried to reduce the detection time and maximize the power of predicting platinum resistance, access, and diagnosis of the tumor in ovarian cancer, using treatment programs and cloud-based systems.
Nwagwu, et al. [64] have assessed the non-clinical endpoint using a digital vivarium cloud paradigm in an ovarian cancer xenograft design. There was a strong relationship between the translation of pre-clinical data and the validness of the parameters for evaluating the illness development and endpoint in mouse oncology cases. A digital cloud paradigm would offer an objective and non-invasive technology enabling real-time evaluation of numerous metrics like movement, breathing, and activity. They have determined the motion loss post-induction and motion loss from the endpoint and compared them with the traditional metrics regarding forecasting illness endpoint. The motion loss post-induction has had a meaningful association with the endpoint after graphed on a linear regression plot. The traditional parameters have had insignificant associations with the endpoint. They have also found that the endpoint's motion loss was meaningfully higher than when the initial medical symptoms to the endpoint have appeared. The outcomes have shown their better capability in forecasting the endpoint than the traditional parameters; before the documented medical symptoms.
Zehra, et al. [63] have provided a short overview of three types of cancer genomics datasets transformed from raw formats into a set of linked datasets within the Linked Open Data (LOD) cloud. The three genomics datasets (Copy Number Variation (CNV), Methylation, & Gene Expression) have been related to ovarian cancer studies and archived initially in three different repositories: The TCGA, Catalogue of Somatic Mutations in Cancer (COSMIC), and CNV in Disease (CNVD)). They have provided these three genomics datasets as a set called LOD for CG (LOD4CG), of five interlinked publicly-accessible SPARQL endpoints that will help researchers and practitioners to explore these datasets and links across them.
Also, Isoviita, et al. [65] have developed a cloud-based ML system (CLOBNET), with the streamlined collection and real-time analysis of rich clinical data derived from the electronic health records, and used the CLOBNET to forecast platinum resistance in patients with High-Grade Serous Ovarian Cancer (HGSOC). They have shown the viability of accessing clinical data from live electronic health records and using ML to analyze it over traditionally-restricting barriers caused by manual steps and interfaces between different organizations and systems. Thus, they have shown that the automated ML approaches such as the CLOBNET are promising as clinical tools in predicting challenging outcomes such as platinum resistance in the HGSOC.
Latip, et al. [66] have tried to enhance the healthcare offerings to patients who have cancer. The O-CareCloud for patient managing mechanisms on the cloud enables access to online health archives of the patients having ovarian cancer to recover and save data and manage them. Their proposed mechanism has emphasized assisting the clinicians in managing and detecting the tumor from an ultrasound test to distinguish a fluid-filled cyst, solid tumor masses, and healthy tissues. Also, the mechanism detects using the outcome of CA (Cancer Antigen)-125 assay blood test measuring the level of CA-125, a tumor indicator, in the blood. • Main aim: Assessing the non-clinical endpoint using a digital vivarium cloud paradigm in an ovarian cancer xenograft design.
• Advantage: High assessing disease progression, Improving the ability of a digital platform to predict • Disadvantage: Failure to perform experiments on humans [63] • Main aim: Providing a short overview of three types of cancer genomics datasets transformed from raw formats into a set of linked datasets within the LOD cloud.
• Advantage: Presenting open data infrastructures

[65]
• Main aim: Developing a cloud-based machine learning system (CLOBNET) to predict platinum resistance in high-grade serous ovarian cancer.
• Advantage: High power predicts platinum resistance

[66]
• Main aim: Enhancing the healthcare offerings to patients who have cancer..
• Advantage: High-quality service, Improving the access and diagnose of the tumor

D. BREAST CANCER
This section reviews the articles that utilize cloud computing in breast cancer treatment. Breast cancer has increasingly become a crucial worldwide health problem since its occurrence is growing, particularly in developing states [124,125]. It appears in breast cells grown oddly and reproduced, creating a lump or tumor [126]. Its initial phase is non-invasive, limited to the breast's ducts or lobules, and not extended to the healthy tissues (named in situ carcinoma) [127]. The invasive breast cancer extends past the ducts or lobules to the breast's healthy tissues or even more to the lymph nodes or other organs. Breast cancer has the highest rank among the cancers regarding the women's mortality rate. It mainly appears in postmenopausal women above 50 years old [128].
However, it appears in men, too, although quite infrequently (about 1% of all breast cancer instances) [129]. Breast cancer is a prevalent type of cancer in developing states. Scholars have usually diagnosed it at the lateral phases. The cancer diagnosis at later stages leads to pain and suffering and brings so many expenses for the caregivers [76]. Today, the prevalent breast cancer diagnosis method is X-ray mammography because of its easiness, portability, and low expenses [74]. We can quicken the decrease in breast cancer mortality by increasing access to top inhibition, on-time diagnosis, and therapeutic services to all women [130]. In this section, 9 breast cancer articles related to cloud-based systems were analyzed. Then the main features of the articles were summarized in Table 6. As can be seen, the researchers have tried to reduce the detection time, make the diagnostic process automated and standardized, maximize the accuracy, promote the image enhancement methods, and enhance the characteristic mining and selection methods in breast cancer using treatment programs cloud-based systems.
Lahoura, et al. [78] have examined cloud computing-based framework for breast cancer diagnosis using extreme learning machine. Saba, et al. [22] have introduced an outline using a cloud-oriented DSS to diagnose and categorize malevolent cells in breast cancer employing breast cytology images. They have used shape-oriented characteristics for the diagnosis of tumor cells. They have also investigated the grading of the affected cells, useful for the grade level needed clinical processes for the patients in the course of the detection. They have also used a cross-validation method to assess the categorization exactness, indicating 98 percent exactness compared to the physical techniques incorporated by a pathologist for the diagnosis and categorization of the malevolent cells. The outcomes have suggested that the introduced method has meaningfully enhanced the diagnosis and categorization of the malevolent cells in breast cytology images.
Agarwal [72] has proposed a web-oriented multiplatform answer for improving the predictive tactics for detecting breast cancer out of diverse examinations like histology, mammography, cytopathology, and fine-needle aspiration cytology, all automatically. The corresponding application incorporates tensororiented data illustrations and deep learning architectural algorithms to generate optimized designs for forecasting innovative examples against each of these clinical trials. He has proposed this mechanism to combine all its calculations effortlessly into a medical environment, not to affect the doctors' efficiency or workflow, instead of improving his competencies. This system can make the detection procedure automated, consistent, quicker, and more precise than the existing standards reached by both pathologists and radiologists, making it helpful from a medical viewpoint to detect expertly with minimal resources. This method can effectively diagnose breast cancer in a standard and optimal way, using the computational intuition of the artificial intelligence algorithms, as confirmed by the precisions, losses, and associations processed on each design's confirmation parameters. It can diagnose patterns far too intricate to be detected by even skilled radiologists and pathologists.
Furthermore, Jiang, et al. [73] have designed and realized a model for distant detection mechanisms in mammography, according to the cloud paradigm. They have used technologies like clinical image information construction, cloud infrastructure, and human-machine detection design to make this model. In contrast, they have designed the storing mechanism using the Hadoop distributed file system technology, allowing users to effortlessly create and activate huge data utilizations and make the most of the cloud computing; its high effectiveness, its scalability, and its cost-effectiveness. Also, the CAD has been conducted by the MapReduce frame. Its detection segment has done the algorithms of fusion of the machine and human intelligence. They have particularly integrated the outcomes of detections extracted from the clinicians' experience and those out of the old CAD, using the man-machine intelligent fusion design, based on the alpha-integration and multi-agent algorithm. The proposed mechanism will be useful for the adjusted health resource, better costeffectiveness, and enhancement of the detection precision in primary health centers.
Mulimani and Kulkarni [74] have proposed expanding cloud in the detection of breast cancer. Cloud computing offers a common pool of assets like storing areas for data, networks, computer computing power, and expert usages for organizations and users. The introduced design has mainly aimed at a cloudoriented DSS for monitoring breast cancer through digital mammograms. They can use the machine in a private cloud as a service. The integration of image enhancement methods, characteristic mining and characteristic selection ones, the collective of neural networks for the categorization, outcomes' confirmation procedure, and application in the private cloud has been useful for the efficiency of the mechanism.
Also, Kao, et al. [75] have introduced an Internetoriented mechanism for calculating the longitudinal variations in the QoL and a cloud-oriented mechanism for managing patients after breast-conserving operation. They have investigated 657 breast cancer patients treated at three tertiary scientific centers. They have indicated that all the mentioned patients had meaningfully better Quality of Life Questionnaire (QLQ-C30) and its additional breast cancer measure (QLQ-BR23) subscale scores during the 2-year follow-up (p<0.05). In the research, the QoL commonly had a negative relationship with older age, high Charlson comorbidity index point, tumor grade III or IV, prior chemotherapy, and extensive postoperative Length of Stay (LoS). However, the QoL had a positive relationship with prior radiotherapy and hormone therapy. Also, having high points for the preoperative QoL inclined to have high points for the QLQ-C30, QLQ-BR23, and SF-36 subscales. According to the feasibility examination outcomes, the 5 blocks have been rated on a Likert scale from 1-7: The system usefulness, ease-of-use, information quality, interface quality, and overall satisfaction.
Also, Bhat, et al. [76] have developed useful tools for physicians to detect breast cancer on-time. They have assessed the adaptive resonance theory to detect breast cancer, using Wisconsin as the dataset. They have done many examinations on a different amount of training and examining datasets and found that assuming the vigilance as 0.5 and the ratio of data as 90 percent for training and 10 percent for the examination has led to improved outcomes. They have shown how Art 1 network is useful in the categorization of breast cancer.
Aruna, et al. [77] have proposed a cloud-based breast cancer DSS for detecting breast cancer through digital mammograms. They have used the proposed mechanism in a private cloud computing setting and conduct it to guarantee data security and back beginners and specialists. After that, the mammograms have been divided into areas, and their characteristics have been derived out of the desired area. They have applied the selected features to the categorization using collective neural networks. They have confirmed the outcomes and presented the reports. • Main aim: Examining cloud computing-based framework for breast cancer diagnosis using extreme learning machine.
• Advantage: High accuracy, Improving recall, Improving precision [71] • Main aim: Proposing a web application showing the cancer prediction outcomes and offering them to clinicians.
• Advantage: High accuracy

[22]
• Main aim: Introducing an outline using a cloud-oriented DSS to diagnose and categorize malevolent cells in breast cancer employing breast cytology images.
• Advantage: High-performance accuracy, Improving detection, Generalization of results is limited

[72]
• Main aim: Proposing a web-oriented multi-platform answer for improving the predictive tactics for detecting breast cancer out of diverse examinations. • Main aim: Introducing an Internet-oriented mechanism for calculating the longitudinal variations in the QoL and a cloudoriented mechanism for managing patients after breast-conserving operation.
• Advantage: High system usefulness, High ease-of-use, High information quality, High interface quality, High overall satisfaction

[76]
• Main aim: Developing useful tools for physicians to detect breast cancer on-time.
• Advantage: Early detection, Remote detection

[77]
• Main aim: Proposing a cloud-based breast cancer DSS for detecting breast cancer through digital mammograms.
• Advantage: High diagnosing, High ensures data security, Supporting both novice and expert users

E. SKIN CANCER
This section has reviewed the selected skin cancerrelated articles where cloud computing played an important role. Skin cancer, a dermatological illness, appears due to the occupational ultraviolet irradiation affecting a substantial number of employees, particularly in agriculture and construction [131]. Skin cancer is a prevalent type of cancer worldwide, and its frequency is growing. In spite of low frequency, the possibility of melanoma metastasize to other parts of the body shows that it is the cause of up to 75 percent of the mortality due to skin cancer. 5-year survival can be as high as 91-95 percent for melanoma if it is detected in primary stages. So, on-time diagnosis and treatment have positive effects on survival. The risk of metastatic spread is lower in the cutaneous Squamous Cell Carcinomas (SCC) [132]. Skin cancer diagnostics is a clinical field where on-time detection enables a better survival rate. Usually, dermatologists detect skin cancer. The number of deaths (due to skin cancer) is high, as there is a limited number of dermatologists [133]. By introducing cost-effective and user-friendly detection tools, scholars make possible the early detection of skin cancer for the primary care doctors and enable monitoring a lot more patients. CuPhysicians currentlyave many tools available ofto offerkin cancer detection [134]. They may process the skin images locally and have restricted detecting abilities; a few transfer the images to the dermatologists for manual analysis to better detect quality. So, we can see an absence of detection quality or high response duration. Skin cancer malignancy is a fatal type of cancer-based on EU statistics. Altogether, it has up to 95 percent chance for treatment in case of on-time detection. The on-time detection is difficult due to low accessibility to expert dermatologists and a low number of patients taking frequent detections [70]. In this section, four skin cancer articles related to cloud-based systems were analyzed. After reviewing the main features, the authors have summarized them in Table 7. As can be seen, the researchers have tried to reduce the detection time, maximize the accuracy, and improve the classification of the dermal cell images in skin cancer using treatment programs and cloud-based systems. Kadampur  Akar, et al. [68] have developed a cloud-based skin cancer diagnosis system using CNN. Their design is cloud-oriented. They have done the detection using a 2-phase CNN pipeline in which a primary CNN is in charge of quality check on the demands and detection CNN whose precision is close to that of the dermatologists. The outputs possibilities over seven diverse lesion groups, consistent with the groups in the international standard industrial classification 2018 dataset, have been employed for training. For training, they have used transfer learning in a ResNet50 network trained on the ImageNet competition dataset.
Osipovs, et al. [69] have introduced a distributed cloud-oriented mechanism to employ the newest detection algorithms for skin cancer and achieve a quick automated detection mechanism. They can process the images using the MATLAB algorithms [135] the skin cancer research group is employing. So, it is necessary to adopt each algorithm to a certain structure of the detection tool. Also, the introduced mechanism relates to several skin analyses of each patient. It has applied central load-balancing server that receives detection demands and transfers image processing demand to the MATLAB processing station. If there is a high load, the balancing server can activate an extra processing station. So, it has the key benefits of a cloud system, including effective resource consumption and quick adopting of the existing requirements, by raising the computing power.
Bliznuks, et al. [70] have implemented portable automated diagnostic devices available to a wide range of medical institutions to deal with the early diagnostics unavailability. They have focused on image segmentation methods and problems and extend portable imaging device flexibility by using cloudbased solutions. They have worked on image segmentation techniques, challenges, and improving a mobile imaging device's adaptability, using cloud-oriented answers. The introduced image segmentation techniques have had positive effects on spectral images. They can be applied to the automated diagnosis of skin cancer. • Main aim: Proposing a non-programming background tool to develop complex deep learning models to categorize dermal cell images and diagnose skin cancer.
• Advantage: Classifying dermal cell images, Detecting skin cancer, High accuracy

[68]
• Main aim: Developing a cloud-based skin cancer diagnosis system using CNN.
• Advantage: Raising in diagnosis power, Raising the speed of processing submitted lesion images [69] • Main aim: Introducing a distributed cloud-oriented mechanism to employ the newest detection algorithms for skin cancer and achieve a quick automated detection mechanism.
• Advantage: Non-invasive skin cancer high diagnostics • Main aim : Implementing portable automated diagnostic devices available to a wide range of medical institutions to deal with the early diagnostics unavailability.
• Advantage: High detecting threats

F. PROSTATE CANCER
In this section, the selected articles that investigated the cloud's role in prostate cancer treatment are reviewed. Prostate cancer is a prevalent kind of cancer among men, the 2nd top reason for death, particularly in developed states [136]. Prostate cancer is the top kind of cancer in older men (> 70) across the world. It is a key health issue, particularly in developed states, as they have a bigger percentage of aging men. Age and family background are the key risk factors. The clinicians can detect it by prostate biopsy in patients having irregularities in their prostate-specific antigen levels or digital rectal exam [137]. The researchers have not yet found all the parameters determining the risk of emerging clinical prostate cancer, though they have recognized some of them [138]. One can name 3 known risk factors for prostate cancer: rising age, ethnicity, and inheritance. If there is a background of prostate cancer in the person's close relative, at a minimum, the risk is twofold. If there is more than one background, the risk grows to 5-11 times [139,140]. In this section, the authors have analyzed four prostate cancer articles related to cloud-based systems. After reviewing the main features, we have summarized them in Table 8. As can be seen, the researchers have tried to reduce the detection time, improve the predictive analysis for cancer growth, increase the collaboration to design, and develop clinical trial data management in prostate cancer using treatment programs and cloud-based systems.
Eguzo, et al. [79] have studied the application of cloud computing and the role of social media for prostate cancer support in Nigeria. The results have shown that cloud computing allows having 1 spokesman; it resolves the necessity to find a spokesperson for each company. Audience-appreciated shortened videos have been applied to describe the illness procedure and the necessity for modified on-time diagnosis. Facebook live presentation has gained the attention the most, with the majority of comments indicating that individuals liked the interference. The reviews have indicated that adding support cellphone video by a survivor has assisted in clarifying prostate cancer.
Gangwal, et al. [80] have surveyed big data predictive examination for diagnosing prostate cancer on a cloudoriented design and getting the taste of a cloudoriented device. Medicine-Radiation Therapy (DICOM-RT) directories having an automatic average atlas-oriented electron density image and quick pelvic organ contouring from the overall pelvis MR scans. They have used the cloud setting as there was a necessity to raise the scalability, functionality, and extensibility, decrease the total cost of the ownership and give access to powerful Central Processing Unit (CPU) and graphics processing unit clusters for practical parallel approaches in clinical image analysis. They have used the cloud structure to show a former technique operating on a cloud interface and generate DICOM-RT files ready for integrating into a commercial treatment planning mechanism.
Wang [82] has presented a model of a cloud-based prototype of a proactive surveillance system for prostate cancer. Using cloud design, diverse cooperating parties can easily design and create medical experimental data management. Even though the Google App Engine (GAE), Amazon Web Services (AWS), and Microsoft Windows Azure can assure high dependability and accessibility, there have still been outages at cloud computing providers in the past. All these cloud computing paradigms employ exclusive Application Programming Interfaces (API), making interoperability more challenging. The results have indicated that cloud computing paradigms like the GAE are appropriate solutions for the upcoming multi-center medical tests. • Main aim: Studying the application of cloud computing and the role of social media for prostate cancer support in Nigeria.
• Advantage: Enhancing learning and retention, Content developing-diagnosis of cancer [80] • Main aim: Surveying big data predictive examination for diagnosing prostate cancer on a cloud-oriented design and getting the taste of a cloud-oriented device.
• Advantage: Predictive analysis for the cancer growth
• Advantage: Decreasing the overall cost, Clinical image analysis [82] • Main aim: Presenting a model of a cloud-based prototype of a proactive surveillance system for prostate cancer.
• Advantage: Increasing in collaborating to design and develop Clinical trial data management

G. GENERAL STUDY
In this section, we have reviewed surveys and articles that investigated the cloud's role in treating any cancer. The detection is important for the effective treatment of the patients [141]. Also, precise cancer categorization is a difficult challenge as the microarray test offers numerous features but a small number of samples [83]. New technologies play an important role in diagnosing diseases. The classifier mechanism has generally applied to cancer categorization to assist the specialists in making a correct detection. The DNA microarray for cancer detection has recently become a famous research field [142]. The main features of each article were summarized in Table 9. The results have shown that the detection and classification of treatment methods are problematic in the field of cancer. So, only skilled experts should do them. The computer-oriented devices should be combined with the identification, diagnosis, and interpretation procedures for that purpose. But, computer-assisted detection encounters the challenge of not having sufficient shared clinical reference data for the introduction, examination, and assessment of processing techniques.
Anuradha, et al. [91] have focusing on one such application in the field of IoT together with cloud computing. Encryption has been done on the blood results of cancer-affected patient and stored it in the cloud for quick reference through the Internet for the doctor or healthcare nurse to handle the patient data secretly. They have provided a framework to enhance the performance of the existing health care industry across the globe. Encryption and decryption has been done using advanced encryption standard algorithm in order to provide authentication and security in handling cancer patients. The task completion time has been greatly reduced from 400 to 160 by using VMs. CloudSim have given an adaptable simulation structure that empowers displaying and reproduced results.
Kečo, et al. [83] have proposed a method for cancer categorization. They have extended it into a parallel algorithm dispersedly computing the microarray data using the Hadoop MapReduce outline to enhance the introduced algorithm's functionality. They have tested the presented algorithm on 11 GEMS datasets and achieved 100% accuracy for less than 25 selected features. Also, its scalability was unrestricted due to the basic Hadoop MapReduce. The outcomes have shown that the introduced algorithm can be useful for the cloud's actual microarray data. Also, the Hadoop MapReduce outline has shown a significant reduction in the processing duration.
Sadhasivam, et al. [84] have surveyed cancer detection epigenomics scientific workflow set up in the cloud computing, using an Improved Particle Swarm Optimization (IPSO) algorithm. They have used the IPSO to achieve appropriate resources and assign epigenomics jobs to minimize the overall expenses of the diagnosis of epigenetic irregularities of probable application in the field of cancer. The outcomes have indicated that the IPSO-oriented job of resource mapping has decreased the overall cost by 6.83% compared to the old PSO. The outcomes of different cancer detections have indicated that the IPSOoriented job of resource mapping can reach lower expenses in comparison to the PSO-oriented mapping for epigenomes scientific application workflow.
Also, Kou, et al. [78] have achieved a cancer prediction model based on mobile healthcare. They have first collected the data from the smart gadgets.
After that, they have sent the data to the server and used the proposed technique at the server to forecast the illness. The main parameters have been recognized, and the predictive categorization has been done, by combining the PCA and SVM as preprocessing phase and post-processing phase, respectively. In the test, a dataset has been created by smart gadgets. The size of the feature subgroup generated by the introduced technique has been decreased by 99.3 percent. Precision, sensitivity, and specificity were assessment pointers; the introduced technique has achieved 82 percent, 82.5 percent, and 81.7 percent, correspondingly. The introduced technique has outperformed the other ones regarding the 3 pointers.The statistical data has improved their confidence in the introduced technique.
Cheng, et al. [90] have developed a cloud-oriented histology database for a cooperative study of cancer. The application of a virtual slide database on a scalable IT architecture will enhance histology data observing and exploring, leading to better worldwide scientific cooperation. They have created the HistoWiz Viewer, the first electronic viewer on the cloud powered by AWS, using a HyperText Markup Language (HTML)oriented deep zoom viewer. It enables the users to immediately observe their virtual slides on any internet-connected PC, Mac, tablet, or smartphone without the necessity to download the virtual slides or install the software or plug-ins. In addition, using AWS saves users thousands of dollars by avoiding the expenses of purchasing and maintaining an in-house IT storage system. Furthermore, Thiebes, et al. [86] have employed the technique used by Nickerson et al. to propose a taxonomy of genome datasets to help the scholars decide whether to keep and process their genome data in the cloud. Their taxonomy has 10 aspects: (1) organism, (2) access, (3) identifiable, (4) file size, (5) processing necessities, (6) transfer necessities, (7) mutable, (8) API access, (9) software accessibility, and (10) the application limitation. The examination of their taxonomy and dataset categorizations from a cloud computing perspective shows the variety of the parameters and contextual impacts beyond just privacy and security issues that can encourage or discourage cancer genomics scholars from sending their genome data to the cloud.
Zhang, et al. [87] have proposed an intelligent cancer prediction technique based on mobile cloud computing. They have first used the primary component analysis to achieve representative characteristics. After that, a simplified characteristic subgroup has been used on SVM based on the sigmoid kernel function. The introduced technique has decreased the size of the produced subgroup by 99.3 percent. It has integrated the cloud's high processing power to the huge storage capacity to make it better and more suitable. They have used the SVM based on the sigmoid kernel function to allocate the suspected cancers in the patients and healthy individuals. The results have shown that the presented model has outperformed the others regarding precision, sensitivity, and specificity.
Kaushik, et al. [88] have suggested a method to create a cancer cloud platform including collaborative tools, security permissions, data harmonization, and made the data easier to query, using metadata curation, resource description frameworks, and visual means. Additionally, they have implemented the common workflow language, an emerging standard for describing computational workflows, to support computational reproducibility. In addition to the motivation, inception, and development of the CGC, they have also presented a case study on applying unsupervised learning methods to identify individual cell types within tumors, using the RNA Sequencing data from TCGA cohorts. They have demonstrated how these computationally-intensive methods make the most of the cloud and how researchers can apply open pipelines to interrogate cancer subtypes and mixed cell populations from the TCGA data on their data. • Main aim: Proposing taxonomy of genome datasets to help the scholars decide whether to keep and process their genome data in the cloud.
• Advantage: Increasing storage and processing of genome data, Increasing access and inspect most controlled-access data sets [87] • Main aim: Proposing an intelligent cancer prediction technique based on mobile cloud computing.
• Advantage: Cancer faster prediction

[88]
• Main aim: Suggesting a method to create a cancer cloud platform including collaborative tools, security permissions, data harmonization, and made the data easier to query.
• Advantage: High security permissions, High data harmonization

V. DISCUSSION AND RESULTS
This study studied and summarized the effects of cloud-based systems on cancer-related studies' success. Accordingly, the scientific articles published until Sep 2021, were reviewed. Consequently, the chosen papers were classified based on the publication year, aim, outcomes, and findings. This study has identified six cancers the cloud affected; colon cancer, lung cancer, ovarian cancer, breast cancer, skin cancer, prostate cancer, and general survey (It has explored cloud computing in cancer generally, not specifically). Table 10 summarizes each article's important features, in which the tick mark denotes that the approaches have tested the desired factors and the cross mark denotes that the approaches have not been tested. As can be seen, the researchers have tried to focus more on the early detection of the sickness, increasing the accuracy and open data infrastructures. The results have shown that most cloud computing studies have been done about breast and colon cancers. We have also concluded that the most common types of cancer in women are breast and ovarian cancers, and the most common one among men is prostate cancer. Therefore, the researchers have tried to develop cloud computing applications to help diagnose diseases faster and improve diagnosis and treatment accuracy [143]. Also, the results have shown that cloud computing is helpful for remotely checking the patients. It is reinforced by the cloud's virtual unrestricted capabilities and resources to remove the technical restrictions such as memory, processing power, etc. Also, the IoT centriccloud design is useful for introducing novel applications and facilities in healthcare [144,145]. Combined with advancements in cloud computing, telemedicine, and artificial intelligence, the ondemand IoT electrocardiograms sensor can potentially help high-risk patients reduce prehospital delays and seek timely life-saving interventions [146]. Also, the studies have shown the most prevalent types of cancer likely to be detected in men and women in 2020; the prostate, lung, and bronchus, and CRCs comprising 43 percent of all instances in men; the prostate cancer alone comprising over 1 in 5 new cases. For women, the three most prevalent types of cancer are breast, lung, and colorectal, including 50 percent of all new cases; breast cancer alone comprises 30 percent of all cases in women [120].
Also, authors have observed that highperformance cloud computing has combined high-performance processing with cloud computing to propose an informatics architecture that delivers the supercomputing power needed for computing and bringing high-quality, multidimensional diagnostic images PCs or mobile gadgets [57,147]. Also, presently, the SVM is an efficient image categorization device, particularly in clinical imaging and traditional medicine [148,149]. Characteristic selection and metric optimization help enhance the SVM outcomes [59]. Fig. 4 shows the number of selected articles in each category that are studied cancer using cloud computing. As can be seen, the highest number of studies in the field of cancer and cloud was related to breast cancer.

VI. OPEN ISSUES AND FUTURE WORKS
We have also found that cancer research is producing a massive amount of data in heterogeneous formats and repositories. Scholars have forecasted that 2-40 exabytes of storing capacity will be required by 2025 exclusively for the human genomes, which will continuously grow nearly 40 petabytes of further genomic information per year [150]. Therefore, these data's heterogeneous nature and their widespread distribution over numerous databases make searching and pattern discovery a tedious and cumbersome task [151]. From a researcher's perspective, a network of coherent and well-interlinked datasets opens the possibilities of advanced search and analysis across such datasets sources to identify novel and meaningful correlations and mechanisms. On the other hand, studying cancer has a strong relationship with histopathology analysis for detection, prognosis, and healing. Nevertheless, most of the important histology data is kept in private documentations on glass slides and hard drives; they don't share them like bioinformatics data. Lack of centrally-located data leads to repetitive, redundant research and makes it difficult for cancer researchers to work collaboratively by limiting widespread access and image sharing. The introduction of a virtual slide database on a scalable IT architecture will help cancer histology data observe and explore, leading to better worldwide scientific teamwork against cancer [152]. The cloud service provider aggregates the consumers' needs for computational power to minimize the total resource requirements, and the overall efficiency is optimized [153,154]. While resources are automatically provisioned and released just-in-time, there is only minimal to no need for auxiliary management effort or service provider interaction. On the other hand, blockchain-based security can also be used to increase security and privacy in medical systems [155,156].
On the other hand, the national cancer institute has recently started searching for chances to work with molecular data in cloud environments. Considering this, scholars can use their tools to work on the data and so prevent huge data sending. In the last 10 years, public cancer collections have significantly contributed to recognizing genomic, transcriptomic, proteomic, and epigenome parameters affecting tumor beginning, development, and healing reactions. Scholars can hire virtual machines in these settings and use processing tools on the data, not requiring data transference to or from anywhere else [157]. This design aims to accelerate the procedure of scientific discovery, decrease the hurdles to entry, and publicize access to the data [158]. So, with the appearance of next-generation sequencing technology, modern medical datasets have much important genomic information associated with a variety of illnesses like cancer. Physicians should investigate, control, keep, visualize, and combine this data to make it medically helpful. Numerous physicians and scholars required to understand these datasets have no IT skills and need efficient and simple mechanisms to work with [116].
Dynamic thermography holds excellent promise as an adjunct modality for breast cancer detection. Many authors are continuing their research and exploring various experimental, clinical and numerical studies to improve the efficacy of this method. However, there are many additional factors to consider before thermography can be used for detection. For example, This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and content may change prior to final publication. numerical simulations provide insight into the thermal interactions within the breast. They also provide a way to study different factors such as metabolic activity, tumor position and depth of tumors within the breast.
Although numerical simulations provide valuable information on the thermal field resulting from the presence of a cancer, the actual breast shape and other factors such as blood perfusion are challenging to estimate. Therefore, simplified breast models have been used. A validated model with the exact breast shape is needed in the literature. This model may be used to predict more accurately the surface temperature distribution of the female breast [159].
The innate difficulties of the traditional healthcare industry are small physical storage, security and privacy, and medical errors [160,161]. As records of the patients contain very important data which should be protected all the time. Existing system is facing lot of inconsistencies in protecting patient's data. The big data has many security challenges and it is increased when it is related to the IoT medical data. Big data management security is the major issue in cloud computing, which usually focuses on data classification and encryption mechanisms [162,163]. It occupies more memory space to store the medical data, which is not cost-effective. Cloud provides high security in order to protect patient's data [164]. The storage costs around multiple times lesser than the server storage, equipment materials, and preparing of HR to keep up and bolster the framework in day by day activities. Since the patient's information is available in the cloud, patient can easily retrieve the prescription from the cloud anywhere any time [91]. According to Fig. 5, when intruders try to illegally see and steal the data, the intruders must jailbreak the security system beforehand. Thus the system will record the attempt and track it down and send the notification to the user using this method, the user will know the who, when, how and where the intruders attempt to attack the system. Finally, nowadays, scholars have extraordinary access to human genomic and proteomic data, quickly improving our present understanding of the complete coverage. Integrating genomic information to shotgun proteomics is still a problem because of the huge increase in the proteomics exploration area [166]. The appearance of next-generation sequencing has changed our capability to produce genomic data. Nowadays, scholars have access to petabytes of multidimensional information from numerous patients. Nevertheless, it makes the investigation of this information more difficult since the bulk of data is increasing [167]. Moreover, the huge bulk of wholegenome sequencing data have been offered by big consortium plans like the TCGA, COSMIC, etc., making unbelievable chances for studying the functional gene and uncovering the cancer-related system. Although the current web servers are useful and prevalent, we should investigate numerous wholegenome analysis functions that are necessary for trial biologists [168].

VII. LIMITATIONS
Authors have limited the search to Scopus, Google Scholar, MEDLINE, Embase, CiteSeer Library, ScienceDirect, ISI Web of Science, IEEE Xplore, PsycINFO, Emerald, and ACM Portal. It is possible that other academic journals provide a good picture of the related articles, too. This study has only reviewed the articles extracted based on keywords such as "cloud" AND "cancer". Cloud computing in cancer might not have been published with those specific keywords. Also, there is a necessity for more pieces of research using other methodologies. Lastly, non-English publications have been ignored. There could be some related papers published in languages other than English.

CONCLUSION
This study has presented an SLR about the roles of cloud-based systems in the success of cancer-related studies. Six important types of cancer have been surveyed that cloud computing is the most widely used in their diagnosis and treatment. We have also illustrated and debated open issues using an exhaustive analysis of 38 essential papers among the basic 142 papers obtained from the search automatically. The results have also demonstrated that most researchers have studied breast cancer.
On the other hand, authors have found that cloud computing eases data protection, privacy, and medical record access. Also, the implementation of cloud computing improves the efficiency and accuracy of diagnosis. Cancer is an important disease across the world, mostly diagnosed in the lateral phases. Its diagnosis in later phases will lead to pain and suffering for the patient and the caregivers' exhaustive expenses. So, cloud computing can detect this disease using new devices at an early stage. The outcomes have also indicated that the cloud-super-based virtual system with motion-oriented navigation enhances the workflow and effectiveness of understanding virtual images for cancer screening. So, cloud computing improves diagnosis, accuracy, treatment, and image resolution and therefore facilitates medical work.

APPENDIX 1
Abbreviation table

Conflict of Interest:
The authors declare no conflict of interest.
Data Availability statement: All data are reported in the paper.